An AI-assisted, failure mode-based toolkit for proactive risk management in radiotherapy: A feasibility study

Area of Science:

• Medical Physics
• Radiotherapy Safety
• Artificial Intelligence in Healthcare

Background:

• Proactive risk management is crucial for safe radiotherapy (RT).
• Formal methods like Failure Modes and Effects Analysis (FMEA) face implementation challenges due to time and resource constraints.
• There is a need for efficient tools to support RT safety activities.

Purpose of the Study:

• To develop and evaluate the feasibility of i-SART (Intelligent Safety Assistant for Radiotherapy).
• i-SART is an AI-assisted toolkit designed to support proactive risk management in RT.
• The study aimed to assess the usability and perceived value of this novel toolkit.

Main Methods:

• Developed i-SART, a web-based prototype with a failure mode (FM) database, submission portal, and conversational assistant.
• The assistant was configured for structured FM analyses, mitigation suggestions, and incident examples.
• Feasibility was evaluated through internal testing for accuracy and an online survey of RT professionals for usability and perceived value.

Main Results:

• The i-SART database comprised 419 FMs across the RT workflow.
• The AI assistant provided clinically plausible mitigation suggestions but had partial accuracy, particularly with guideline references and incident details.
• Survey results from 51 professionals showed favorable median ratings (4/5) for ease of use, risk management support, and potential error reduction.

Conclusions:

• The failure mode-based toolkit (i-SART) is technically achievable and positively received by early users for risk management.
• Further development is needed to expand the FM database and improve the AI's accuracy, addressing hallucinations.
• Future research should evaluate i-SART's impact on risk assessment processes and patient safety outcomes.